Video models

Select a model with the model field of POST /videos/generation. All models currently support text-to-video.

Models

Model	Resolutions	Duration (seconds)
`seedance-2.5`	`480p`, `720p`, `1080p`	4–30
`seedance-2`	`480p`, `720p`, `1080p`, `4k`	4–15
`seedance-2-fast`	`480p`, `720p`	4–15
`seedance-2-mini`	`480p`, `720p`	4–15

Input fields

Field	Type	Default	Description
`prompt`	string	—	Required. What the video shows.
`seconds`	integer	`4`	Video duration. Must be within the model’s range.
`definition`	string	`480p`	Output resolution. Must be supported by the model.
`orientation`	string	`9:16`	Aspect ratio: `21:9`, `16:9`, `4:3`, `1:1`, `3:4`, `9:16`, or `auto`.
`watermark`	boolean	—	Whether to add a watermark.
`seed`	string	—	Seed for more repeatable results.

Any other field returns 400 invalid_parameter. The defaults are the cheapest options, so set seconds and definition explicitly when you need more.

{
  "model": "seedance-2.5",
  "input": {
    "prompt": "A product shot of a ceramic mug rotating on a marble table",
    "seconds": 8,
    "definition": "720p",
    "orientation": "9:16"
  },
  "max_cost_usd": "2.00"
}

Pricing

You pay per second of output video. The rate depends on the model and resolution: Cost = rate per second × seconds See current rates on the pricing page in the API console. Each job keeps the rate that applied when it was accepted, and cost.amount_usd on the job shows the exact charge.

Capping the cost

Set max_cost_usd to the most you are willing to pay for a job, as a decimal string with up to six decimal places. If the job would cost more, the request fails with 400 max_cost_exceeded and you are not charged. This protects you from unexpected charges if prices change.

Getting started

Core concepts

Models

Models

Input fields

Pricing

Capping the cost

​Models

​Input fields

​Pricing

​Capping the cost

Models

Input fields

Pricing

Capping the cost